Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
Python agentaianthropicclaude-codecompressioncontext-engineeringcontext-windowcursorfastapilangchainllmmcpopenaiprompt-engineeringproxypythonragtoken-optimizationtokenstypescript
style: apply ruff format to _strip_per_call_annotations
Garm0eb6efd55892
diff --git a/headroom/proxy/helpers.py b/headroom/proxy/helpers.pyindex 12c943cbe..58b3b837f 100644--- a/headroom/proxy/helpers.py+++ b/headroom/proxy/helpers.py@@ -293,11 +293,7 @@ def _strip_per_call_annotations(obj: Any) -> Any:breaking prompt-level aggregation downstream."""if isinstance(obj, dict):- return {- k: _strip_per_call_annotations(v)- for k, v in obj.items()- if k != "cache_control"- }+ return {k: _strip_per_call_annotations(v) for k, v in obj.items() if k != "cache_control"}if isinstance(obj, list):return [_strip_per_call_annotations(item) for item in obj]return obj@@ -360,9 +356,7 @@ def compute_turn_id(elif system is not None:try:normalized_system = _strip_per_call_annotations(system)- h.update(- json.dumps(normalized_system, sort_keys=True, default=str).encode("utf-8")- )+ h.update(json.dumps(normalized_system, sort_keys=True, default=str).encode("utf-8"))except (TypeError, ValueError):passh.update(b"\0")